On Mon, Mar 19, 2012 at 04:46:42PM -0400, Paul Novotny wrote:
> On Sun, 2012-03-04 at 13:09 -0600, Steve M. Robbins wrote:
> > Other architectures show a regression in that now the libraries don't
> > even build whereas before they did build. Still others claim a
> > dependency installability problem for minc or gdcm.
> >
> > [1] https://buildd.debian.org/status/package.php?p=insighttoolkit4&suite=experimental
>
>
> Here is a one line patch that should help ITK get past the current build
> errors on big-endian systems (I think). How do you test on these
> architectures? VMs? This probably isn't necessary in the long run, since
> I think you are going to use the system TIFF.
>
> -Paul
> diff --git a/debian/patches/big_endian.patch b/debian/patches/big_endian.patch
> new file mode 100644
> index 0000000..2b2d90c
> --- /dev/null
> +++ b/debian/patches/big_endian.patch
> @@ -0,0 +1,11 @@
> +--- a/Modules/ThirdParty/TIFF/src/itktiff/tif_predict.c
> ++++ b/Modules/ThirdParty/TIFF/src/itktiff/tif_predict.c
> +@@ -560,7 +560,7 @@
> + for (count = 0; count < wc; count++) {
> + uint32 byte;
> + for (byte = 0; byte < bps; byte++) {
> +- #if WORDS_BIGENDIAN
> ++ #ifdef WORDS_BIGENDIAN
I don't understand this change. On the big endian systems,
surely something is #defining WORDS_BIGENDIAN. What is
it defined as and where is the define? I'd be surprised
if the #define was no to "1", in which case #if and #ifdef
are both true, so what does this change accomplish?
Thanks,
-Steve
Attachment:
signature.asc
Description: Digital signature